Changeset 235

Timestamp:

1999-04-12T15:45:05+12:00 (25 years ago)

Author:

rjmcnab

Message:

Finished the query filter.

Location:

trunk/gsdl/src

Files:

• colservr/filter.cpp (modified) (2 diffs)
• colservr/filter.h (modified) (1 diff)
• colservr/queryfilter.cpp (modified) (3 diffs)
• colservr/queryinfo.h (modified) (1 diff)
• recpt/Makefile (modified) (2 diffs)
• recpt/librarymain.cpp (modified) (2 diffs)

trunk/gsdl/src/colservr/filter.cpp

@@ -12 +12 @@
 /*
    $Log$
+   Revision 1.4  1999/04/12 03:45:02  rjmcnab
+   Finished the query filter.
+
    Revision 1.3  1999/04/06 22:20:30  rjmcnab
    Got browsefilter working.
@@ -220 +223 @@
   set1 = resultset;
 }
+
+
+// tests to see if el is in set
+bool in_set (const text_tset &set1, const text_t &el) {
+  text_t::const_iterator here = el.begin();
+  text_t::const_iterator end = el.begin();
+  text_t tryel, tryel_add;
+  bool first = true;
+
+  // the element is in the set if any of its parents are
+  // in the set
+  do {
+    // get next possible element to try
+    here = getdelimitstr (here, end, '.', tryel_add);
+    if (!first) tryel += ".";
+    first = false;
+    tryel += tryel_add;
+
+    // see if this element is in the set
+    if (set1.find(tryel) != set1.end()) return true;
+  } while (here != end);
+
+  return false;
+}

trunk/gsdl/src/colservr/filter.h

@@ -113 +113 @@
 void intersect (text_tset &set1, const text_tset &set2);
 
+// tests to see if el is in set
+bool in_set (const text_tset &set1, const text_t &el);
+
 #endif

trunk/gsdl/src/colservr/queryfilter.cpp

@@ -12 +12 @@
 /*
    $Log$
+   Revision 1.2  1999/04/12 03:45:03  rjmcnab
+   Finished the query filter.
+
    Revision 1.1  1999/04/06 22:22:09  rjmcnab
    Initial revision.
@@ -20 +23 @@
 #include "queryfilter.h"
 #include "fileutil.h"
+#include "queryinfo.h"
+
+
+// some useful functions
+
+// translate will return true if successful
+static bool translate (gdbmclass *gdbmptr, int docnum, text_t &trans_OID) {
+  infodbclass info;
+
+  trans_OID.clear();
+
+  // get the info
+  if (gdbmptr == NULL) return false;
+  if (!gdbmptr->getinfo(docnum, info)) return false;
+
+  // translate
+  if (info["section"].empty()) return false;
+
+  trans_OID = info["section"];
+  return true;
+}
+
 
 
@@ -163 +188 @@
 }
 
-void queryfilterclass::filter (const FilterRequest_t &/*request*/,
+void queryfilterclass::filter (const FilterRequest_t &request,
                 FilterResponse_t &response,
-                comerror_t &err, ostream &/*logout*/) {
-  response.clear();
+                comerror_t &err, ostream &logout) {
+  outconvertclass text_t2ascii;
+
+  response.clear ();
   err = noError;
-}
+  if (gdbmptr == NULL) {
+    // most likely a configuration problem
+    logout << text_t2ascii 
+       << "configuration error: queryfilter contains a null gdbmclass\n\n";
+    err = configurationError;
+    return;
+  }
+  if (mgsearchptr == NULL) {
+    // most likely a configuration problem
+    logout << text_t2ascii 
+       << "configuration error: queryfilter contains a null mgsearchclass\n\n";
+    err = configurationError;
+    return;
+  }
+
+  // open the database
+  gdbmptr->setlogout(&logout);
+  if (!gdbmptr->opendatabase (gdbm_filename)) {
+    // most likely a system problem (we have already checked that the
+    // gdbm database exists)
+    logout << text_t2ascii
+       << "system problem: open on gdbm database \""
+       << gdbm_filename << "\" failed\n\n";
+    err = systemProblem;
+    return;
+  }
+
+  // get the query parameters
+  int startresults = filterOptions["StartResults"].defaultValue.getint();
+  int endresults = filterOptions["EndResults"].defaultValue.getint();
+  text_t index = filterOptions["Index"].defaultValue;
+  text_t subcollection = filterOptions["Subcollection"].defaultValue;
+  text_t language = filterOptions["Language"].defaultValue;
+  queryparamclass queryparams;
+  queryparams.collection = collection;
+  queryparams.search_type = (filterOptions["QueryType"].defaultValue == "ranked");
+  queryparams.casefolding = (filterOptions["Casefold"].defaultValue == "true");
+  queryparams.stemming = (filterOptions["Stem"].defaultValue == "true");
+
+  OptionValue_tarray::const_iterator options_here = request.filterOptions.begin();
+  OptionValue_tarray::const_iterator options_end = request.filterOptions.end();
+  while (options_here != options_end) {
+    if ((*options_here).name == "StartResults") {
+      startresults = (*options_here).value.getint();
+    } else if ((*options_here).name == "EndResults") {
+      endresults = (*options_here).value.getint();
+    } else if ((*options_here).name == "QueryType") {
+      queryparams.search_type = ((*options_here).value == "ranked");
+    } else if ((*options_here).name == "Term") {
+      queryparams.querystring = (*options_here).value;
+    } else if ((*options_here).name == "Casefold") {
+      queryparams.casefolding = ((*options_here).value == "true");
+    } else if ((*options_here).name == "Stem") {
+      queryparams.stemming = ((*options_here).value == "true");
+    } else if ((*options_here).name == "Index") {
+      index = (*options_here).value;
+    } else if ((*options_here).name == "Subcollection") {
+      subcollection = (*options_here).value;
+    } else if ((*options_here).name == "Language") {
+      language = (*options_here).value;
+    } else {
+      logout << text_t2ascii
+         << "warning: unknown queryfilter option \""
+         << (*options_here).name
+         << "\" ignored.\n\n";
+    }
+
+    options_here++;
+  }
+
+  queryparams.search_index = index+subcollection+language;
+  queryparams.maxdocs = (endresults > 100) ? endresults : 100;
+
+  // do query
+  queryresultsclass queryresults;
+  mgsearchptr->setcollectdir (collectdir);
+  if (!mgsearchptr->search(queryparams, queryresults)) {
+    // most likely a system problem
+    logout << text_t2ascii
+       << "system problem: could not do search with mg for index \""
+       << queryparams.search_index << "\".\n\n";
+    err = systemProblem;
+    return;
+  }
+
+  // assemble document results
+  if ((request.filterResultOptions & FROID) || (request.filterResultOptions & FRranking) ||
+      (request.filterResultOptions & FRmetadata)) {
+    int resultnum = 1;
+    ResultDocInfo_t resultdoc;
+    text_t trans_OID;
+    vector<docresultclass>::iterator docs_here = queryresults.docs.begin();
+    vector<docresultclass>::iterator docs_end = queryresults.docs.end();
+    
+    while (docs_here != docs_end) {
+      if (resultnum > endresults) break;
+      
+      // translate the document number
+      if (!translate(gdbmptr, (*docs_here).docnum, trans_OID)) {
+    logout << text_t2ascii
+           << "warning: could not translate mg document number \""
+           << (*docs_here).docnum << "\"to OID.\n\n";
+    
+      } else {
+    // see if it is in the set (or the set is empty)
+    if (request.docSet.empty() || in_set(request.docSet, trans_OID)) {
+      if (resultnum >= startresults) {
+        // add this document
+        resultdoc.OID = trans_OID;
+        resultdoc.ranking = (int)((*docs_here).docweight * 10000.0 + 0.5);
+        response.docInfo.push_back (resultdoc);
+      }
+      
+      resultnum++;
+    }
+      }
+      
+      docs_here++;
+    }
+  }
+
+  // assemble the term results
+  if ((request.filterResultOptions & FRtermFreq) || (request.filterResultOptions & FRmatchTerms)) {
+    queryresults.sortqueryterms();
+    queryresults.uniqqueryterms();
+
+    TermInfo_t terminfo;
+    bool terms_first = true;
+    vector<termfreqclass>::iterator terms_here = queryresults.terms.begin();
+    vector<termfreqclass>::iterator terms_end = queryresults.terms.end();
+
+    while (terms_here != terms_end) {
+      terminfo.clear();
+      terminfo.term = (*terms_here).termstr;
+      terminfo.freq = (*terms_here).termfreq;
+      if (terms_first) terminfo.matchTerms = queryresults.termvariants;
+      terms_first = false;
+
+      response.termInfo.push_back (terminfo);
+
+      terms_here++;
+    }
+  }
+
+  response.numDocs = response.docInfo.size();
+  response.isApprox = (queryresults.getnumdocs() != queryparams.maxdocs);
+}

trunk/gsdl/src/colservr/queryinfo.h

@@ -102 +102 @@
   vector<docresultclass> docs;
   vector<termfreqclass> terms;
-  vector<text_t> termvariants;
+  text_tarray termvariants;
 
   void clear ();

trunk/gsdl/src/recpt/Makefile

@@ -12 +12 @@
 CC = gcc
 CCFLAGS = -g -Wall -Wunused -pedantic -W -Woverloaded-virtual # -Wshadow
-DEFS = -DNZDL -DQUIET -DSHORT_SUFFIX -DPARADOCNUM -DUSE_FASTCGI -DGSDLSERVER
+DEFS = -DNZDL -DQUIET -DSHORT_SUFFIX -DPARADOCNUM -DUSE_FASTCGI -DGSDLSERVER \
+       -D_LITTLE_ENDIAN
 INCLUDES = -I../../lib -I../../packages/mg-1.3d -I../../packages/mg-1.3d/lib \
            -I../../packages/mg-1.3d/src/text -I../../packages/fcgi/include
@@ -96 +97 @@
 
 install:
-#   cp recpt /home/nzdl/nzdl-1.2/cgi-bin/rrecpt; \
-#   cp library /home/nzdl/nzdl-1.2/cgi-bin/rlibrary
-    cp $(EXEC) /home/nzdl/nzdl-1.2/cgi-bin
+    cp recpt /home/nzdl/nzdl-1.2/cgi-bin/rrecpt; \
+    cp library /home/nzdl/nzdl-1.2/cgi-bin/rlibrary
+#   cp $(EXEC) /home/nzdl/nzdl-1.2/cgi-bin
 
 depend:

trunk/gsdl/src/recpt/librarymain.cpp

@@ -12 +12 @@
 /*
    $Log$
+   Revision 1.4  1999/04/12 03:45:05  rjmcnab
+   Finished the query filter.
+
    Revision 1.3  1999/04/06 22:20:35  rjmcnab
    Got browsefilter working.
@@ -70 +73 @@
   recpt.add_protocol (&nproto);
 
-  cgiwrapper (recpt, "cnrub");
+  cgiwrapper (recpt, "gberg");
   return 0;
 }